摘要:
An improved method and apparatus for buffering and issuing instructions for use with superscalar microprocessors are disclosed. The method comprises the steps of: (a) obtaining an instruction buffer comprising a plurality of entries, each entry comprising a random access memory (RAM) portion and a content addressable memory (CAM) portion for storing result data and source operand tag, respectively, wherein the CAM portion also contains means for linking with an associated RAM portion and the result data contains an instruction; (b) providing a result bus capable of transmitting the result data and a result tag; (c) matching the result tag in the result bus with the source operand tag in the CAM, and writing the result data into the RAM portion of an entry if the result tag in the result bus matches the source operand tag of an associated CAM portion; and (d) issuing ready instructions and changing the source operand tag in a corresponding CAM in such a manner that the entry containing the CAM will be identified as an empty entry so as to all new instruction to be written thereto. Because instructions are stored in the RAM in an out-of-order, a linear systolic array is provided so as to keep the sequence of instructions in order. The linear systolic array, which can be easily compressed, allows the prioritization of instructions for issue among ready instructions, and the handling branch mis-prediction and faults to be implemented.
摘要:
A process is provided for determining the beginning and ending of each instruction of a variable length instruction. Data lines are stored in a first memory area which illustratively is an instruction cache. Each data line comprises a sequence of data words that are stored at sequential address in a main memory. The data lines contain multiple encoded variable length instructions that are contiguously stored in the main memory. Multiple indicators are stored in a second memory area, including one indicator associated with each data word of the data lines stored in the first memory area. Each indicator indicates whether or not its associated data word is the initial data word of a variable length instruction. A sequence of data words may be fetched from the cache. The fetched sequence of data words includes a starting data word and at least the number of data words in the longest permissible instruction. Plural indicators (i.e., a vector of indicators) are also fetched from the second memory area including one indicator associated with each of the data words of the fetched sequence. Using the indicators as delimiters of the sequence of to-be-decoded instructions, one or more non-overlapping subsequences of the sequence of data words are identified, wherein each subsequence of data words is comprised in a different, sequential to-be-decoded instruction. Each subsequence of data words is then decoded as a separate instruction.
摘要:
Methods and systems are disclosed for exploring instruction-level parallelism in superscalar processors by renaming stack entries. In a first embodiment, the stack renaming is implemented in a parallel structure that renames the instructions in parallel. In a second embodiment, the stack renaming is implemented in a serial structure that renames the instructions serially. In a third embodiment, the stack renaming is implemented in a combined parallel-serial structure that renames the instruction partially in parallel and partially in series.